Skip to content
View cuihangbin's full-sized avatar
  • Bytance seedance

Block or report cuihangbin

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
cuihangbin/README.md
Hangbin — FlashInfer kernel engineering, CUDA optimization, SGLang and vLLM inference

Hi, I'm Hangbin 👋

I work on LLM inference infrastructure, with a focus on GPU kernels, attention backends, and quantization. My open-source work centers on SGLang, FlashInfer, vLLM, and SageAttention.

./current_work

Area Current work
Frameworks SGLang attention DP routing and load accounting; vLLM attention backend integration, FP8 quantization correctness, DeepGEMM integration, and Model Runner V2 profiling.
Kernels FlashInfer Gated DeltaNet prefill on SM120; SM90 FA3 FP8 paged attention; SageAttention sequence-length handling and paged KV-cache decode for CUDA Graphs.

./selected_systems

Frameworks

Project Contribution Status
SGLang Count the unprefilled portion of chunked requests in total_tokens routing load. PR #50088 · Development branch · Reproducible benchmarks. Open
vLLM #55732 — SAGE_ATTN backend integration Open
vLLM #52753 — Pin the static FP8 quantization pattern to group_shape=None Open
vLLM #49928 — Keep DeepGEMM FP8 quantization inside an opaque op for TMA scales Open

Kernels

Project Contribution Status
FlashInfer #6342 — Experimental SM120 CuTe GDN prefill backend Merged
SageAttention #402 — SM89/SM120 sequence lengths and paged KV decode for CUDA Graphs Open
sgl-flash-attn Recover SM90 FA3 FP8 paged hdim256 performance with chunked two-level P@V accumulation while retaining the precision fix. PR #55 · Development branch · Integration into sgl-kernel. Merged

Validation and performance results are documented in the linked PRs and apply to their tested hardware and workloads.

Pinned Loading

  1. sgl-project/sglang sgl-project/sglang Public

    SGLang is a high-performance serving framework for large language models and multimodal models.

    Python 38.9k 10.7k

  2. flashinfer-ai/flashinfer flashinfer-ai/flashinfer Public

    FlashInfer: Kernel Library for LLM Serving

    Cuda 7k 1.8k